Google admits mistakes with Gemini AI; co-founder says more testing needed after controversial images
-
Google co-founder Sergey Brin admitted the company "messed up" on the AI image generator Gemini, which produced controversial "woke" images.
-
Brin said the mistakes were "mostly due to not thorough testing" and upset many people. Google paused Gemini in the wake of the fiasco.
-
Brin defended Gemini, saying rival chatbots also say "weird things" that "feel far-left." He said any AI model can be prompted to generate questionable content.
-
Since the controversy, Brin said Gemini has improved by 80% in producing more accurate images.
-
Google CEO Sundar Pichai previously apologized and said the tool would sometimes "overcompensate" in seeking diversity where it didn't make sense.